Gecchimia n Cosmochimia Aaa Vol. 50, pp. 2267-2279 Q Rrpmon lourods Ltd. 1986. Printed in U.S.A.

cml6-7037/86/$3.00 + .oo

Rare earth element patterns in Archean high-grade metasediments and their tectonic significance

STUART Ross TAYLOR’, ROBERTA L. RUDNICK’, Sco’rr M. MCLENNAN’ and KENNETH A. ERIKSSON~

‘Research School of Earth Sciences, Australkn National University, Canberra, ACf 2601 Australia ‘Department of Geological Sciences, Virginia Polytechnic Institute, Blacksburg, VA 2406 I, U.S.A.

(Received April 3, 1986; accepted in revised&m July 3, 1986)

Abstract-Metasediments from two contrasting types of Archean high-grade terrains are interpreted as being derived from distinct tectonic settings. The Kapuskasing Structural Zone, Canada, represents the deep roots of a typical greenstone belt, whereas the Limpopo Province, southern Africa and Western Gneiss Terrain, Australia, mainly consist of shelf sediments deposited on a granitic basement and then buried to the depths required for granulite fties metamorphism. Upper amphibolite to gramtlite facies paragneisses from the Kapuskasing Structural Zone have REE patterns similar to those of greenstone belt sediments, except where partial melting has oamned, forming restites with Eu enrichment and melts with Eu depletion. Except in this latter instance, metamorphism has not atfected the primary REE patterns. REE patterns in Atchean upper amphibolite to gram& tkies metasediments from the central Limpopo Province and Western Gneiss Terrain show wide diffemnces, ranging from patterns which resemble those in post-Archean tenigenous sediments, to typical Archean sedimentary rock patterns. The diversity in REE patterns for these shallow shelf metasediments is interpreted as resulting from derivation horn local provenances, Samples with “post- Archean” patterns, displaying Eu depletion, ate inkrpmted as being derived from K-rich granitic plutons which were portions of small, stable early Archean terrains, pmcurso rs of the widespread late Atchean- Proterozoic episode of major cratonic development.

CONTENTS preted to indicate that the processes of erosion, trans- portation and deposition provide thorough mixing and

REE AS INDlCATOBS OF CRUSTAL EVOLUTION Post-Archean sedimentary rocks

averaging of these differing source patterns. Thus the

Archean sedimentary rocks and the significance of the REE patterns preserved in the sedimentary record

Archean-Proterozoic boundary provide an overall average of the composition of the Archean high-grade terrains upper crustal rocks exposed to erosion.

GEoLOGICAL sETTINcs Kapuskasing Structural Zone (KSZ), Canada The Limpopo Province, Southern Ajiica The Western Gneiss Terrain (WG f). Western Australia

ANALY’IKAL RESULTS Sample selection Methoak Rare earth element data Other data

INTERPRBTATION OF ANALYTICAL RESULTS Metamorphic effkcts The role of accessory phases Provenance

TECIDNK IMPLICATIONS Recycling and the previous extent of high-grade terrains A model for Archean crustal growth

REE AS INDICATORS OF CRUSTAL EVOLUTION

RARE BARTH ELEMENTS (REE) are relatively insoluble elements, with short residence times in the ocean. Ac- cordingly, they are transferred almost quantitatively from upper crustal sources to terrigenous sedimentary rocks. The observed uniformity of REE patterns in post-Archean tenigenous sedimentary rocks, in con- trast to the diversity of source rock patterns, is inter-

Post-Archean sedimentary rocks

Post-Archean terrigenous sediments display a de- pletion in Eu, relative to the other REE, when nor- malized to chondritic RRE abundances. Since mantle derived igneous rocks (the ultimate source ofthe con- tinental crust) rarely display either relative depletion or enrichment in Eu and since no upper crustal res- ervoir enriched in Eu has been identified, the depletion of Eu in the upper crust has been ascribed to intra- crustal processes, whereby K-rich felsic ‘igneous rocks, depleted in Eu, are produced. Petrologica! and isotopic evidence favour intracrustal melting, rather than frac- tional crystallization, to product large volumes of K- rich granitic rocks (e.g. WYLLIE, 1977; ALL&% and BEN OTHMAN, 1980). In this model, Et&+ is retained in plagioclase, in the lower or middle region of the crust (TAYLOR and MCLENNAN, 1985). Such reservoirs have not been positively identified. The ubiquitous de- pletion of Eu in upper crustal tenigenous sediments is thus considered due to the domination of the present composition of the upper crust by K-rich granites, granodiori@s and other felsic igneous rocks produced by intracrustal melting.

2267

2268 S. R. Taylor er a/

Arch&m ~~d~~entar~ rocks and the signi~~~nce qf the Archem-Proterozoic boundary,

Studies of REE patterns in low-grade Arcbean sed- imentary rocks from Western Australia, South Africa, Canada, and Greenland reveal a significant difference from that observed in the post-Archean record (see recent review by MCLENNAN and TAYLOR, 1984). There is more diversity in the patterns, which may be flat, MGRR-like, or steeply tREEenriched, typical of Na-rich plutonic rocks (tonalites, trondhjemites, with occasional Eu enrichment) and felsic volcanics. These patterns are attributed to local provenance effects. Mostly the patterns in the sedimentary rocks are in- termediate, approximating 1: 1 mixtures of the extreme basic and felsic patterns. Significant Eu anomalies are absent. This evidence is consistent with an Archean upper crust dominated by the bimodal basa.h:tonalite- trondhjemitedacite igneous suite. Archean sedimen- tary rocks sampled are typically located in greenstone belts, but their petrography and chemistry indicate that they sample a wider provenance. For example, the rne~~rnen~ at Kambalda, Localized in a basic-ul- trabasic volcanic sequence, contain a substantial Na- granitic component (BAVINTON and TAYLCIR, 1980). Since the presence of 10% or more of K-rich felsic igneous rocks with Eu depletion would be r&xted in a corresponding depletion of Eu in the sediments, the general absence of such ef&cts has been taken as evi-

dence that K-rich granites and granodiorites were scarce (< 10% exposure) in the upper crustal rocks being sam- pled. These observations have led to a model of crustal growth in which the Archean upper continental crust is dominated by the bimodal basalt:tonalite-trondh- jemite-dacite suite, mostly of mantle derivation, whereas the post-Archean upper crust is dominated by K-rich granites, granodiorites and associated volcanic rocks, the products of intracrustal melting (TAYLOR and MCLE~AN, 1985).

Major intracrustal melting events, preceded on time scaks of IO* years by a massive addition of materiai from the mantle, are thus thought to occur over a pe- riod of 500-700 Ma towards the end of the Archean. This event is proposed to constitute the major episode in the growth of the continent crust, about 75% hav- ing been formed by about 2500 Ma ago, at the com- mencement of the Proterozoic Era.

Archean high-grade terrains

Archean sedimentary rocks found in high-grade me~mo~hic terrains have received littie attention in formulating these models, partly because little gee- chemical work has been carried out in such regions. The limited REE data that have been reported from Greenland (BOAK e[ al., 1982; DYMEK et al.. 1983; MCLENNAN et al., 1984), Montana-Wyoming (MUELLER et al,, 1982; GIBBS et al., 1986) and India (NAQVI et al., 1983) indicate considerable variabihty, suggestive of the dominating influence of local prove- nance effects. There is growing evidence that maw

hip-DDE terrains contain me~im~nta~ rocks deposited in a different tectonic setting from those 01 the well studied greenstone belts.

In this paper, we study REE data for me~dirne~” tary rocks from two contrasting types of Archean high- grade terrains. These are: 1) the Kapuskasing Structural Zone of Central Ontario, Canada, and 2) the Limpopo Province of southern Africa and Western Gneiss Ter- rain of Western Australia. The former region consti- tutes a high-grade ~uiv~ent of typical iate Archean greenstone belts, whereas the latter regions are exam- ples of high-grade areas containing sedimentary rocks deposited in a cratonic environment.

GEOLOGICAL SETTINGS

Rocks of the Kapuskasing Structural Zone (Fii 1) represent highly metamorphosed equivalents of the widespread low- grade Archean granite-greenstone terrains. The KSZ transects the Superior Structural Province of emmaI Ontario, extending from the eastern shores of Lake Superior northeast to the southern shores of James Bay. Along the southern portion, in the Chap~u-Foleyet region, granuhte facies gneisses are in fault contact with rocks of the Abitibi greenstone belt to the east and grade into lower grade rocks of the Michipicoten greenstone belt to the west, over a distance of about 120 km. This gradation from gram&e through green&S thcies recks is interpreted to represent an obhque cross section through about 20 km of Arcbean crust (PERCIVAL and CARD, 1983). Detailed zircon g~hronol~c work by PERCIVAL and KROGH (1983) reveals the fol~o~ng tectono-thermal history prior to overthmsting deposition of the vokanic-sedimentary sequence of the Wawa and Abitibi greenstone belts and in- trusion of plutons - 2750-2696 Ma ago, metamorphism and tectonism 2700-2685 Ma ago. and post teetonic plutonism 2685-2668 Ma ago.

Pressures and temperatures of metamorphic equilibmtion increase gradually from west to east across the KS2 PHtClVAt. (1983)~two~-gnadem~orphiczones.(f)nK garnet-clinopyroxene-plagioclase zone, characterizing the transition from amphibolite to granulite facie& extends from the Ivanhoe Lake cataclastic zone in the east, weshvard into tonalitic rocks of the Wawa subprovince and (2) The ortho- pyroxene zone, which characterizes the granuhte fties, OXXLIS in four patches within the eastern and northern portion of the KSZ. Mineral pair the~obaromet~ yields temperatures > 800°C in the east, ranging to under 600°C in the west, with pressures ranging from 5.4 to 8.4 kbar (PERCIVAL, 1983).

in contrast with the Ksz, the Limpopo Province (Fig. I) forms an example of those Archean hijJh-@ade terrains which consist predominantly of metasedimentary rocks, and tack the extensive metavolcanic sequences typical of greenstone terrains. The Limpopo Province is bounded to the north and south by typical Archean granite-greenstone terrains. The m~imen~ rocks studied here, part of the Beit Bridge Complex, overlie Saud River gneisses (3786 i: 61 Ma jRb- Sr), BARTON er a/., 1983), and are intruded by the Messma layered intrusion (3270 + 105 Ma (PbPb) &G?TON ef i& 1983). The Beit Bridge supracmstals are thus e&y Archean in age. Metamorphic pmasures and temperatures am estimated at about 10 kbars and over 800°C (HORROCKS, 1983). LithObgiG3l associations present in the Beit Bridge Complex

have been studied by ERIKSSON and RIDD f I985}, Qwutzit~ up to 100 m thick are interpreted to be of detrital sedimentary origin, on account of high concentrations of heavy minerals (zircon. rutile}, rather than as recrystallized cherts (FRIPP,

Archean BEE patterns

285 30%

22%

4PN

I kletamdlment

Fe/sic Metaigneous

m Amdmsite

2269

26%0’S

FIG. 1. Generahxed geological maps and sample locahties for the Limpopo Province (top left) (atter WATKEYS, 1983), the Kapuskasiing Structural Zone (bottom left) (atIer PERCIVAL, 1983) and the Mount Narryer ngion of the Western Gneiss Terrain (r-i&t) (after WILLIAMS ef al., 1983).

1983). They are interlayered in places with amphibolites, in- mrpmted as basic sills and dikes. W&spread biotite+rnet- cordieri~sillimanite 8neimes resemble shales in composition and are inmrpmted as metapelitee (BRANDL, 1983; PRIPP, 1983). Associated makmetite quart&s (metacherts) mpresent metamorphosed iron formation. Units of marble and talc- silicate 8neiss up to 30 m thick are inmrpmted as metasedi- mentary carbonates or calcareous-pelites. The protolith of the fourth association, the 8ray 8neisses, is inmrpmted as arkose (ERIKSSON and ICJDD, 1985). These authors have estimated the litholo8ic propottions of the primary sedimentary rock types in the Belt Bridge Complex as: mudstone: 50%; quartz arenite: 17%; arkose: 17%; and limestone and calcamous mudstone: 17%.

Such a sequence is clearly different from the Archean 8men9tonebelttenainsandrepmaentsadistincteariyArchean tectonic setting. It resembles the wideapmad quartz arenite- carbonate association which characterizes shallow cratonic shelf environments in the Proteroxoic and Phaneroxoic. The Beit Bridge supmcrus@ are thus considered to constitute an Archean precursor of later widespread cratonic shelves.

The Western Gneiss Terrain (WGT), Western Australia

The Western Gneiss Terrain is situated along the western- most mar@ of the Archean Yii Block, Western Austmha. This terrain is compoaed mainly ofbaaded. locally migmaiitic, granitic (s.1.) Bneips with interlayered metasedimentary and mafic htholopies (GEE et I& 198 1). Tonaiitic rocks, the dom- inant rock type in many Amhean terrains, are only minor components in this r&on (GEE et al., 1981). Metamorphic grade is 8enemlly upper amphibolite, but 8ramrlite facies as- semb@es am locally developed (e.g., orthopyroxene-garnet assembw in some metapelitea).

The samples invest&ed in this study come from the northernpottionoftheWestemGneissTerrain,inthevicinity of Mount Narryer (Pii 1). Major rock types in this Fcgion are

old (>3380 Ma) adamehitic and syenograaitic sneiases, fras- ments of a disrupted layered ma6c intrusion, and younger metasediments (MYERS and WLLL+MS, 1985; KINNY et al.. 1986). tie the Limpopo meta&iments, the htholo&c as- semblages in the Mt. Narryer retgion are mc of shal- low shelfdepositional environmentz quart&, con8iomerate, pelite, carbonates and banded iron formation (GEE ef al., 1981).

The depositional a8e of the memsediments is not precisely known, but xircon 8eochronology on the Mt. Narryer quartx- ites and metapelites constrsinsittobe.between325OMa(the ageoftheyoungestdetritaldrcons)andeo.2800Ma(tbe~ of metamorphic rims on xircons)(K~~~~, 1986;KRJNyef cl., 1986).Assedimentswouklbeexpectedtocontainasampk of detrital zircons from any exposed &ous tack, the lrvge time interval between 3250 and 2800 Ma shown by zircon ages in the metasediments indicates either a lack of igneous activity during this period or deposition of the sediments soon after3250Ma.Thelatterhypotbisisamsidwdmorclikely.

ANALYTICAL REwLTs

Sample selection

A description of samples and sample localities is &en in the Appendix. It is always dithcub to mco8mxewithcertainty, metapelitic lithologiea in high-grrde metamorphic terrains. Litholo8ical associations, such as inter&red metaquartxites and carbonates, the pmaence of ttraphite, hi& A&Q content or the occurrence of aluminous minerals such as fjamt, sil- hmanite and cot&rite and evidemm for production of the rock throt@r weatherin pmcessm from major element data (e.g.. high Al2QJNafl ratios) ail may be used as ev&nce for a sedimentary protofith. None of these chsrrcteristics alone is diagnostic.

In the Limpopo and WGT suites, sedimentary pamnta8e is Clear on the grounds of mineralo8y, composition and WI- obgic association. A sedimentary or@ for samples horn the

2270 S. R. Taylor m al

LP LP LF LP iLP LP KSZ-l KSZ-a KSZ-II KS-12 KS.?-‘3

MN MN * 4 I, jr. 20 28 14 "5 FN

14.6 29.2 39.8 80.7 8.81

35.7 7.60 1.40

7.09 1.31

10.8

2.97 11.0 14.5

226 0.58

24.6

50.6 4.92

20.5

3.99 1.07

14.0 0.68 41.9

34.1 1.56 73.1 5.16 0.19 9.15

25.9 0.68 29.7 7.30 0.26 5.45 2.75 0.09 1.38 8.29 3.25 I.38 - a.68 0.26 2.5,

I.80 - 0.57 4.94 0.14 1.66

4.43 0.12 1.60

16.0 13." 59.1

32.5 25.2 1'3 3.85 3.06 11.6

15.8 12.2 40.8

3.57 2.35 6.20

1.26 0.94 1.39 4.27 1.78 3.38 0.80 0.28 0.50

5.19 1.61 2.68

‘9.’ 9.05 40.1: 20.5

4.27 2.37

'__II 9.78 2.76 2.02

4,

76 7.12

22.4 2.85 0.43 2.00 0.38 2.61 0.58 1.68 1.56

159

0.53 17.8 14

65.0 6.85

26.3

5.99 1.23

5.51 0.92

5.73

49.7 4.89

17.4 4.16 0.77 4.26

0.75 4.86 0.90 2.66 2.24

123 0.56

a.8

23

16.h 3.!5

1 .o: C.46 1.hP 2.07 0.29 c.44 I .:2 2.97 c.15 0.70 0.92 :.oa

0.89 2.22

9, 5L 1 .5; 0.71

1i.i 2.8 ,i.' :-

3.15 0.42 2.32

1.14 3.38 7.01

0.42 1.21 1.26

115 0.95

1.07 3.02

0.30 0.48

0.83 1.33 0.82 1.28 63 242

I .41 0.93

,0.7 33.0 8.G 9.4

2.94

91 0.99

EREE ;60

EulEu* 0.65 120- 4.2 171 !.07 1.0 I.0

2.1 3.9 17.7

45 1.25 19.5

O.Ub

6.9 LdN/YbN 3.8 Y 29

1.9 13.2

64 15.7 3.7 27 !‘

nathcxl: spark .wuR?e mass spctrmerry (SSt4.s). nay1or and Gortan, 1977) :no data

Kapuskasing Structural Zone is somewhat leas obvious because of the lack of sedimentary structures or associations. The gen- erally high Al,03 content (up to 18.2%) and ram occurrence of graphite do, however, point to a sedimentary origin.

Two samplea, KSZ- I2 and RSZ- 13 are from the same out- crop and, on the basis of field evidence, are clearly atTected by partial melting.

Methods

Rare earth elements, Th, U and some other trace elements were determined by spark soume masa spectmmetry using the method described by TAWR and GoR?oN (1977). Other elements were determined by a combination of elaqon mi- croprobe on fusion glpIpes, X-my thmmaun a: spectrommy,

The Kapuskasing samples, where unmodified by partial melting have REE patterns typical of low-grade Archean sed- iments (Figs. 2,3). Two samples (RSZ-1 and KSZ-6) pceaess variably enriched LREE ((J.a/Yb)N = 26 and 5, respectively), with no Eu anomaly (Fig. 2). The two samples which have undergone partial melting (Fii. 3) have compkmentaty REE pattcms. The restite, RSZ-13, is LREE enriched ((La/Yb)N = 13) with a large positive Eu anomaly and the melt, KSZ- 12, is extremely LREE enriched ((La/Yb)N = 49) with a slight negative Eu anomaly (Eu/Eu* = 0.93). If these two patterns are combined in 5050 proportions, the resultant REE psttem closely matches that of unmelted paragn& KSZ-1 (Fii 2). Pamgne&sampleKSZll isLREEemiched((La/Yb)u- 16) and has a large positive Eu anomaly. These features resemble those of ICSZ-13, the paragneiss restite, suggesting that RSZ- 11 may also be a restite.

The Limpopo me&sediments provide an interesting con- trast. These samples show a wide variation in REE patterns (Table 1, Fig. 4). Two samples have patterns which resemble those of post-Archean tenigenous sediments (PAAS, NASC, ES, TAYLDR and MCLENNAN, 1985; Fig. 4a), with s&n&cant depletion in Eu (Eu/Eu* values are 0.56 and 0.65). A third sample (LP-30) has a peculiar V-shaped pattern (Fii Sk which may be due to garnet concentration through metamorphic di&rentiation. This is discussed in more detail in the fblknving

plasma spe&omet& The d&Is of which &men& wem%e- termined by which methods are provided in the tables of an- alytical data (Tables l-5).

Rare earth element data

The results am given in Table 1 for the six Limpopo and three Western Gneiss T&n metaaediments and five Ra- puskasing paragneisses.

KAPUSKASING

I I , I I I I 1 I I I 1 La Ce PI Nd Sm Eu Gd Tb Dy Ho Er Yb

FIG. 2. chondrite-normahxed RFE plots for metasedimentary rocks from the Kapuskasing Structured Zone. Note that the pattems have variaMe La/Yb ratios and lack negative Eu-anomalies. Sample RSZl 1, with the positive Eu-anomaly, may have had a small partial melt fraction removed. These patterns are similar to those found in sedimentary rocks deposited in most Archean greenstone belts.

2271

KAPUSKASING

l KSZl2 (melt) n KSZ 13 (restite)

01/l Mix

I ’ ’ ’ ’ I I I I I I I I

Lo Ce Pr Nd Sm Eu Gd Tb Dy Ho Er Yb

FIG. 3. Chondrite-normal&l REE plots of melt and restite phases of a single outcrop of pamgneieo from the Kapuskasing Structuml Zone. Also shown is a KEE pattern derived from l/l mixing of the restite and melt phases. This pattern is similar to those shown in Fig. 2 and indicates there probably has not been large scale mmoval of any melt phase on an outcrop scale

section. The other thme Limpopo metasediments display quite different patterns (Fig 4b). !%mple LP-20 has no Eu anomaly and a high La/Yb ratio, msembling the Archean Yellowknife and Pilbam sediments (JENNER et al., 198 1; MCLENNAN ef al., 1983), suggesung it is derived from a source dominated

100 h. LIMPOPO (a)

\ . LP4 4

FI Ill I I I I , I I I

La Ce Pr Nd Sm Eu Gd Tb Dy Ha Er Yb

FIG. 4. Chondritc-normalii KEE plots of metasedimen- tary rocks from the Limpopo Province. a). Samples showing significant negative Eu-anomalies. These KEE patterns are typical of post-Archean sedimentary rocks and dilfer from typical Archean sedimentary rocks found in greenstone belts. Such pattems indicate that in some regions, Archean crust underwent intmcmsml melting to produce K-rich granitic rocks, which were then incorporated into terrigenous sedi- ments. b). Sampks showing variable La/Yb ratios and lacking negative Eu-anomahes. Sample LP-14 is a femtginous me- tachert and is p&ably horn an iron-formation. These data indicate that even in the high-grade ten&s which have sed- imentary rocks pomeasing post_Archean type REE patterns, othersedimentarymcluwithREEpattemstypicalofAtchean gmenstone belta are also common. Ovemll, these data suggest that the K-rich cmtonic sources, with negative Euanomahes, were a local rather than regional feature.

by felsic igneous rocks (MCLENNAN and TA~R, 1984). The KEE pattern of sample LP-28 resembks the patterns of many low-grade Atchean metasediments, with no Eu anomaly, but is much tlatter, consistent with a pmponderance of basaltic source material (MCLENNAN and TAYLDR, 1984). Sample LP-14, a ferruginous metachert, is generally similar to LP-28, although with very low total REE abundances due to dilution with quark. This sample is p&ably part ofan iron-formation sequence. KEE patterns of Archean iron-formations may, in some cases, reflect hydrotheemai inputs (FRYER ef al., 1979) and accordingly am di@icult to interpmt. If cmstal sources am being m&ted in the REE da* the slightly steeper, mom LKEE-entiched pattern, compared to LP-28, would indicate a more fekic provenance, with no signihcant Eu-anomaly.

The three Western Gneiss Terrain samples have compar- atively uniform HKEE patterns (Fii 6). All are chamcterimd by LKEE enrichment and substantial negative Eu anomalies with Eu/Eu* = 0.46-0.71. HKEE patterns are flat at about 6-9 times chondritic levels. Them is &ability in tbe degree of LKEE enrichment with (La/Sm)N - 2.8-9. I; the silica-rich samples (impure metaquart&ea) having the greatest degree of LREE enrichment. These REE paftems, in gener& resemble those in typical post-Archean sedimentary rocks.

Other data

The other geochemical data are presented in Tables 2-5. The major element data are patticukrly interest@ (Tabk 2; Fig. 7). The Kapuskasing samples are all characterized by low Al20,/Na20 ratios and low K2O/Na20 ratios. Such data in- dicate a general absence of significant weathering and/or a Na-rich provenance. These chamcteristics are typical of the majority of Archean ten@ous sedimentary mcks @iCtiN- NAN and TAYLOR, 1984). Apart from the metacheet (LP-14), the Limpopo sampks fag into two distinct groups. One group (LP-20, LP-28) has low Al&/N820 and K20/Na20, similar to the Kapuslcaaing sampks described above. These am also the samples with the typical Archean-like REE patterns. The other group (LP-4, LP- 11, LFKW), with negative Eu m am characterized by high Al~03/NaI0 ratios and K20/Na20 rat& indicative of fairly severe weathming in the soume rocks (NESBITT and YOUNG, 1984). Samples from the Western Gneiss Terrain are similar to this latter gmup but with vatiabk K20/Na20. The low CaO, particularly in the WGT samples, also indicates weathering pmcemes a&ted the source rocks. The data am also consistent with a K-rich soume composition.

Most of the samples analyzed have high Th/l_l ratios (>4) which are normally associated with post-Amhean rrcycled upper crustal sources (e.g. TAYLOR and MCLENNAN. 1985). However, because of mobility of Th and U during high grade metamorphism, and the possibility of aceusov phase control

2272 S. R. Taylor et al

I 1 1 t 1 I I I , 1 I I

La Ce Pr Nd Sm Eu Gd Tb Dy Ho Er Yb

FIG. 5. Chondrite-normalixed REE plot of a Limpopo metapclite showing a negative Eu-anomaly and merit. The HREE-enrichment is heat explained hy addition of garnet during metamorphic

~ntiation, however the small amount of met present in the sample is insufficient to explain the magnitude of the enrichment. A complex metamorphic history may he indicated. In any case, we interpret the negative Eu-anomaly as being a primary feature and accordingly include this sample with the others showing negative Eu-anomalies (see Fig. 4a).

on Th and U abundances in the WGT quart&s (see below), theTh/Uratioscannothcmgardedasaprimaryfeature.

AnumheroftheLimpoposampkaandthemetq&efrom the WGT have high Cr and Ni concentrations (Table 5), with Crvpluesupto57OppmandNivrluernachieg22oppm. Although Amhean shales commonly have elevated Cr and Ni concentratiom related to a m&c component in their prove- nance,anomalouslyhighCrandNilbuadaraceawithasso- ciated high Cr/V (~4) and Ni/Co (~7) ratios have been noted for shales from the F&t Tree and Moodks Group from South Aftica,aswellassomeotherAtcheanmquencm(DANCHIN, 1967; TAYLCBR and MCLEMNAN, 1985). The Limpopo and WGTsamplestendtohanCr/VandNi/Coratioswhichare typical of most shales and accm&gly the elevated Cr and Ni abundances are not anomalous.

INTERRURA’ITON OF ANALYTICAL RESULTS

Metamorphic t&cts

Whole-rock compositions can be a&ted by gran- ulite f&es metamorphism in two ways: 1) by loss of elements to a fluid phase and 2) by partial melting of the protolith. In general, K, Rb, Cs, U, and Th are most suscqtible to depletion during granulite facies metamorphism; the REE are generally considered to

bc immobile (HEIER, 1973; TARNEY and WINDLEY, 1977; RIJDNKK et al., 1985). Granulitcs mostly show U depletion, whereas Th, Rb and possibly K, may or may not be depleted (RUDNKK et al., 1985). In addi- tion, in situ partial melting will partition elements be- tween the melt and rest&e portions of the migmatite, thereby changing the compositions from that of the protolith. However, unless the partial melt is removed from the vicinity, no net compositional change is cre- ated by partial melting.

Of the metasediments investigated here, only some of the KSZ samples have been modified by partial melting. Samples KSZ-12 and KSZ13 represent the melt and restite, respectively, of a partiahy melted par- agneiss outcrop, with biotite remaining as a stable re- sidual phase. When mixed together in 5050 propor- tions, the resultant major and trace element compo- sition closely matches that of unmelted paragneiss KSZ- 1. Thus, partial melting has not resulted in a net compositional change on outcrop scale.

K/Rb ratios for the KSZ paragneisses are similar to those of unmetamorphosed sedimentary rocks (Table 3), suggesting that fractionation of K from Rb has not occurred during metamorphism. Similarly, Cs contents

L”’ I I I I I I I I

La Ce Pr Nd Sm EU Gd Tb Dy Ho tr Yb

FIG. 6. C&mdrite-normali& REE plot of metasedimentary rocks from the Mount Narryer region Of the Westrrn013eigTaPin.MBomplarbowLREEenrichmenfflatHREE~andnceativcEusnomriics. The variable enrichment of LREE my be r&tcd to concentrations of the heavy mineral monazite in the quark-rich samples. Such REE patterns are consistent with a K-rich cratonic provenance.

Archean REE patterns

Table 2 Hajor element deta, in ueisht percent

2213

LP LP 4 11

Limpopo Bait Kepuskasing Structural Zone usstern cne1ss Terral”

LP LP LP LP KS2 KS2 KS2 KSZ KSZ Ia m Ia 30 20 28 14 1 0 11 12 13 45 FN K

SlO. 58.6 57.5 63.0 69.1 56.5 84.5 63.1 62.0 67.3 68.3 59.8 56.0 80.9 84.9

TlO, 0.72 0.37 0.71 0.42 1.60 0.003 0.60 0.79 0.40 0.45 0.68 0.61 0.13 0.03 .u,o,

;;.; . ;;.; 1;.; ‘:.;I 14.0 0.06 17.5 18.2 16.5 15.8 17.4 21.6 12.6 11.6

Fe0 0:39 0:18 0:OSO

14.4 13.8 5.83 5.50 3.30 4.05 6.87 9.10 5.35 1.20

nno 0.14 0.34 0.069 - 0.15 - - - 0.09 0.18 0.15 MO 6.26 3.74 5.40 2.87 5.74 0.55 3.12 4.05 2.16 2.57 3.94 10.4 0.58 2.10 cao 1.05 0.68 1.9 1.70 4.4 0.05 3.59 3.60 4.14 2.18 2.78 0.012 0.10 0.017

Na,O 0.55 0.23 1.16 2.79 1.49 0.05 3.27 3.14 3.50 3.29 3.36 0.41 0.10 0.15 K,o 2.69 2.00 3.65 2.77 0.94 0.01 2.32 2.51 1.59 2.76 3.17 1.39 0.12 0.02

I 99.6 99.2 99.8 100.0 99.4 99.1 99.9 99.9 98.9 99.4 98.0 99.6 100.1 100.2

Methods: Electron microprobe on fusion glasses for Kaspwkasing samples; cceibination electron microprobe. ICP and AA for Limpopo end Western Cnelss samples, - :no data.

are high and K/Cs ratios are comparable to those of sedimentary rocks, suggesting Cs contents were not af- fected by metamorphism. The retention of K, Rb and Cs in the metasediments is consistent with the apparent stability of biotite during the metamorphism. In con- trast with K, Rb and Cs contents, Th/U ratios are high in all but one sample (KSZ-11 Table 4), suggesting sign&ant U depletion relative to Th. These samples have La/Th values ranging from 2.7 to 6.8. Most Ar- chean sedimentary rocks have La/Th near 3.6 jL 0.4 (MCLENNAN et al., 1980). Thus, little Th depletion can be accomodated in these samples. The one sample with low Th/U (KSZll) has a very high La/l% ratio (26, Table 4), suggesting Th depletion. The low Th/U ratio and absolute abundances may reflect retention of these elements in an accessory phase such as zircon.

Three of the Limpopo metasediments and the WGT metapehte have experienced U depletion with little or no change in Th (Th/U ratios > 6 and La/Th ratios of 2.3-3.8). Of the remaining Limpopo and WGT samples, two have low Th/U (LP28 and LP14), with relatively high La/Th, suggestive of depletion of Th. The other samples (LP20, MNFN, MNK) have Th/U and La/Th similar to unmetamorphosed sedimentary rocks. Th and U for the two WGT quartz-rich meta-

sediments (MNFN and MNK) are probably controlled by detrital accessory phases (see next section). It is not known whether LP20 has undergone depletion in Th and U or has retained its origmal LILE concentrations.

The Limpopo sample (LP30) with tire enriched HREE is interpreted as follows: tiom L&id, the pat- tern resembles that ofpost-Archean sedimentary rocks, with a sign&ant depletion in Eu. From Tb to Yb, the pattern rises steeply, enriched in the heavy RFE, typical of garnet (zircon cannot account for this enrichment; see next section). Although the rock contains l-2% garnet, approximately 10% garnet would have to have been added to the rock in order to explain the REE pattern. In addition, the garnet is unaltered and shows no signs of breakdown. Therefore, while the REE re- quire some form of garnet concentration on a hand specimen scale, presumably through metamorphic dif- ferentiation, this is not supported by the petrography. The reasons for this paradox are unclear although ear- lier metamorphic events, not preserved in the present petrography, could account for this discrepancy. Nonetheless, excluding the HREE enrichment, the overall REE pattern of this sample sugge&r a similarity to that of post-Archean tenigenous sediments, with a significant negative Eu anomaly.

Table 3 Trace element data for lerge ion lithophlle elements, in ppm.

Llmpopa Belt Kapuskaslng Structural Zone weetern oneiss terrain

LP LP ifi

LP ii

LP KSZ KSZ KSZ KSZ Ksz m 4

Ia m 11 20 1u 1 8 11 12 13 ‘I5 FN K

ca 3.1 0.48 3.9 11.7 0.14 - Ir.9 2.0 0.8 1.3 3.4 0.4 Kb

0.18 0.19 - _ - _ _ _

65 8::

42 620

89 Be 3w

95 - - - 290 600 170 4.0 1055 495 sr 793 1270 460 :; 32 37 154 31

73 487 130

1.5 449 384 452 Pb 11 la 1163 2.7 2.0 17 3.2 0.34 9.3 6.0

3.0 8.0 K,Rb 12.7 7.6 1.8 8.6 2.5 - - - _ _ _

296 385 314 279 225 - - -

nethods: SCM eXM)Pt Sr in Llmpopo end Yeatern Cneiae samples by ICP and Rb end Sr in KWUekeain.3 semplea by XRF. - :“a data.

‘214 S. R. Taylor et al

1.37 1.45 2.32 1.62 1.91 9.15 12.5 14.8 7.69 1.39 6.7 8.6 6.4 4.7 0.73 2.38 4.55 4.95 4.0 3.22

87 180 183 - 90

36 39 37 i" 2.0 1.6 9.5 3.9 8.0 2.0 0.14 1.7 0.5 1.3 11 9.0 39 II 12

3.18 2.34 2.69 15.2 10.; 0.29 0.96 0.64 0.64 0.036

0.4 0.73 0.36 0.31 G.8, U.32 0.10 6.1, 2.69 0.49 22.1 5.58 2.4 8.4 7.5 1.6 27 17

3.60 4.12 4.84 3.81 3.39 1.3 129 148 158 120 I?,

36 36 33 i? 3b 0.3 2.7 2.2 2.0 3.1 3.. 1.0 , .i 1.6 1.2 I.5 2: 0.58 7.5 1.9 6.6 8.3 ,L.L 6.80 6.86 5.95 26.5 ?.67 3. 4.

I () . 62

The role of accessory phases

Accessory phases can afkct the concentrations of REE, Th and U in sediments if they are preferentially concentrated by sedimentary prooesses during depo- sition. This factor is particularly sign&ant when con- sidering coarser-grained elastic sediments such as sandstones and siltstones. A number of the samples investigated here are quartz-rich, suggestive of being sandstones originally. In addition, zircon, apatite and monazite are present in some of the samples (see Ap pendix), so a briefdiscu&on of their possible influence on the REE patterns and Th and U concentrations is warranted.

Concentration of detrital zircon could potentially cause HREE enrichment in sediments. The amount of zircon present can be deduced from whole rock Zr concentrations (Table 4). The highest Zr content in

our samples is 180 ppm; this corresponds to ~0.04% modal zircon, if all this Zr is in zircon. The HREE content of zircon is extremely variable (Yb = 30 to 4000 ppm; GROMET et al.. 1984). Therefore, detrital zircon could potentially cause a chondrite-normalized increase of 0.05 to 6 in Yb, dependent upon the Yb content of the zircon. However, all terrigenous sedi- ments will contain some amount of detrital zircon as an integral component (post Archean shales typically have 200 ppm Zr; TAYLOR and MCLENNAN, 1985). Therefore, a sediment would have to possess unusually high Zr contents in order for concentration of detrital zircons to influence the HREE content. This is not observed in any of the samples investigated here. It is worth noting that the only sample to show significant HREE enrichment (LP-30) has far too high a Yb value to be contributed by detrital zircon.

Apatite and monazite can also contribute sign&ant

A Western Gnriss Terrain

00 Limpopa / //‘/

s Kopuskasing

increasing effects of / / weaihering

FIG. 7. Plot of Al203 versus Na20 for metasedimentary rocks from Archean high grade terrains. Samples with open symbols have negative Eu anomalies (Eu/Eu* < 0.9) and samples with filled symbols do not (Eu/Eu* z 0.9). Also shown are lines of constant Ai203/Na@ ratio, corresponding to average Arehean and post-Arehean shales (TAYLOR and MCLENNAN, 1985) and Pleistocene Amazon Cone muds, derived fi-om highly weathered sources (KRONBERG ef al.. 1986). The heavy arrow indicates the general trend of increasing effects from weathering. !Snmpies from the Kapuskasing region (lacking negative Eu-anomalies) show little or no signs of weathe&g, whexeas WGT samples (with negative Eu-anomalies) are derived from severely weathered sources. The Limpopo samples show variable influences of weathering, which comlate with negative Eu anomalies.

Archcan REE patterns

Table 5 Trace element data lor ferrmagnesian Waae elements, in ppm.

2275

Limpop-3 Belt Kapuskasing Structural Zone western cneiss TerraIn

LP LP LP LP LP 3": 20 28 14

K.52 KS2 KS2 KS2 KS2 lu rar UN 4 11 1 8 11 12 13 45 FN K

Cl- 517 V 225 W/V 2.3 SC 32 Nl 220 CO 50 NUCo 4.4 CU 100 nn 1000 zn - Li 6

255 260 230 120 2.2 172 - - 142 269 570 05 146 98 250 4.6 130 - - 85 145 260

3.0 1.8 2.3 0.48 0.48 1.3 - - 1.7 1.9 2.2 13 23 12 39 0.2 - - - - - 35 43 110 120 29 5.5 69 - - 73 79 213 18 26

k', 0.6

0% 9.2 _ _ _ _ _ 60

2.4 4.2 - - - - - - 44 7.0 1.4 8.0 47 - - 40 21

139 700 2630 535 _ _ _ _ _ 710 _- - _- 88 - - 86 122 - 42 35 10 - _ _ _ _ _

60 26 27 12 2.5 2.2 5.5 7.8 31 12 12

3.6 11 1360 1160

Methods: ICP TOP Llmpopo Samples and Yestern Gnelss; KRF Ior Kapuskaslng samples. - :"o data

amounts of U, Th and RFE. Lack of PzOs data pro- hibits estimates of the concentrations of these minerals in the samples, but thin section examination shows these phases are volumetrically less than 0.1%. The REE are trace components of apatite, but are major constituents of mom&e. Consequently, the REE pat- terns of the two monazite-bearing WGT quartz-rich metasediments may be controlled by this phase. Mi- croprobe analyses of LREE, Th and U contents of monazites in samples MNK and MNFN are given in Table 6. These data show that only 0.01% monazite in MNK and 0.03% monazite in MNFN are required to account for all the LREE and Th in these samples. Similarly, 0.03% monazite in MNFN could account for all the U in this sample, but at least 0.06% monazite would be required in MNK to account for the uranium content. An unrecognized U-rich accessory phase may supply this additional uranium. Thus? the LREE en- richment and high Th and U contents of quartz-rich samples MNK and MNFN, in comparison to sample MN45, are probably due to accumulation of detrital monaxite.

It is apparent from the discussion in this section that REE abundances in sandstones may be seriously af- fected by the presence of resistant accessory minerals which are concentrated by sedimentary processes. The potential dominance of such secondary effects in silt- stones and sandstones indicate that these sedimentary lithologies may be less reliable indexes of broader crustal evolution than shales. For these reasons, we have mostly concentrated on using shales and similar fine-grained sedimentary rocks in our studies of the

TABLE 6. LREE. Th and U contents of nonazites, wt. percent.

IUK IUIFN

Pmlge Average Range Average n-5 n-8

La 12.3-13.2 12.1 13.2-16.4 14.0 Ce 25.3-26.7 26.1 24.2-31.2 26.0 Nd 7.97-8.48 8.23 7.07-9.08 7.70 Sm 1.51-1.67 1.60 1.65-2.16 1.76 Th 5.62-6.85 6.32 1.78-8.00 5.13 " 0.14-0.37 0.29 1.39-1.96 1.62

evolution of the continental crust (e.g. TAYLOR and MCLENNAN, 1985).

Provenance

In the general absence of alteration of REE patterns due to metamorphism, we interpret the KSZ, Limpopo and WGT REE patterns as primary, reflecting those of the original sedimentary rocks (except for sample LP-30). The chemical characteristics of the KSZ sam- ples are very similar to those of greenstone-belt sedi- ments (MCLENNAN and TAYLOR, 1984). The steep to moderate LREE-emichment and low A120,/Na20 and K20/Na20 ratios of the KSZ samples point to the mixing of variable amounts of relatively unweathered tonalite-trondjhemitedacite and mafic rocks to pro- duce the sedimentary protoliths.

In contrast, the Limpopo samples are characterized by chemical variability. Archean sedimentary rocks display much more diversity in REE patterns than their post-Archean counterparts, consistent with less efficient mixing of Archean source lithologies (e.g., MCLENNAN and TAYLOR, 1984). The Limpopo province represents an extreme example of this diversity, similar in many respects to the REE data from the early Archean high- grade me&sediments from west Greenland (McLEN- NAN et al., 1984).

Three ofthe Limpopo samples (LP14, LP20, LP28) have patterns resembling those of low-grade Archean sedimentary rocks, with provenances ranging from ba- sic to felsic, and without Eu anomalies. Such patterns recall the diversity encountered in the Akilia and Mal- ene supmcmstal succeSSions (MCLENNAN et al., 1984). In this context, the Limpopo metasediments do not differ from those of many other Archean terrains.

The other three samples (LP4, LPI 1, LP30), as well as the three Australian samples, in contrast, show REE patterns typical of those of the post-Archean terrigen- ous sequences (TAYU)R and MCLENNAN, 1985). They must be derived, to a large degree, From weathering and erosion of K-rich granitic plutons. This is consis- tent with derivation from a small-scale cratonic source, analagous to the present-day upper crust. The major element data, with high K20/Na20 ratios, is also con-

2216 S. R. Taylor el al.

sistent with such a source. The observed Eu depletion is thus inherited from a K-rich granitic source, which in itself was derived by partial melting within the crust. This granitic source area could not have been exposed to erosion on a large scale, for the sedimentary rocks bearing the signature of Eu depletion are not ubiqui- tous.

TEaONIC IMPLICATIONS

An important conclusion from these data is that Archean high-grade terrains cannot be grouped to- gether and considered to be a distinctive tectonic as- sociation differing from low-grade terrains. The Ka- puskasing, and no doubt other high-grade terrains, simply represent highly metamorphosed equivalents of greenstone belts. REE data from metasedimentary rocks found in such regions can, accordingly, be ex- pected to have the same characteristics as sediments in greenstone belts.

In the Limpopo Province and Western Gneiss Ter- rain, early Archean metasediments 3.6-3.2 Ga in age preserve evidence of deposition in shallow-shelf envi- ronments. It seems reasonable to suppose that they were deposited at the margins of a small craton (GEE et al., 198 1; ERIKSSON and KIDD, 1986). The REE patterns indicate that a variety of sources were avail- able, with input from basic and felsic sources, including K-rich granites with negative Eu anomalies. Isolated sedimentary basins or environments have preserved this evidence on a scale of hundreds of square kilo- meters. The presence of K-rich granitic rocks contrib- uting Eudepleted REE to sedimentary basins has the following implication. Such Eudepleted source rocks must arise by intracrustal melting at depths not ex- ceeding about 40 km (i.e., within the plagioclase sta- bility field). However, available data suggests that such Eu depleted sediments are uncommon in the Archean record, so that cratonic development and inferred in- tracrustal melting must be very limited, relative to the widespread bimodal igneous activity in Archean ter- rains. The Na-rich tonalite-trondhjemite-dacite asso- ciation, at one end of the bimodal suite, are derived through partial melting of mafic-ultramafic composi- tions from depths where garnet is stable (e.g. BARKER ef al., 1981).

The evidence suggests that localized segments of thr Archean crust underwent this sequence of events. It is curious that only sediments in high-grade terrains pre. serve this record and that the evidence of intracrustal melting is lacking in other Archean terrains. This ob- servation has important tectonic implications. It im- plies separate development of greenstone belts and of cratonic environments preserved in high-grade rerrains. The Eudepleted rocks. sources for the high-grade shelf sediments. were not contributing (< 10%) debris to the sedimentary basins preserved in low-grade terrains (greenstone belts). This appears to contrast with mod- em sedimentary environments; for example. modem deep sea turbidite sands from differing active tectonic settings (fore-arc. back-arc, continental arc) very com- monly show trace element and isotopic evidence for derivation from both old recycled upper continental crust as well as young volcanic-plutonic sources (MCLENNAN et al.. 1985). Rare earth element data for Phanerozoic greywackes show a similar pattern (BHA- TIA, 1985). Only some sediments deposited in fore-arc basins of wholly oceanic island arcs (e.g. modem Mar- ianas Basin; MCLENNAN et al., 1985, Devonian Bald- win Formation; NANCE and TAYLOR, 1977) are devoid of a signature of intra-crustal melting.

Recycling and the prevwus extent of high-grade terrains

There are no detailed compilations of the relative proportions of high-grade versus low-grade supracmsial successions, although such information is critical for models of Archean crustal evolution. Examination of geological maps suggests that high-grade rocks may comprise as much as 5oo/o of all Archean supracrustal exposures. Of these, some proportion is simply a highly metamorphosed equivalent of greenstone belts.

It is thus necessary to postulate the following se- quence of events for the development of the basement and cover rocks in the Limpopo Province and Western Gneiss Terrain:

(i) Formation of thick crust (>20-30 km). (ii) Intracrustal melting to produce Eu depleted fel-

sic rocks.

Preferential preservation of cratonic environments, preserved in high-grade terrains is also likely. It is now well established that intracrustal recycling processes are a dominant factor in the preservation and present- day distribution of crustal rocks and that recycling rates vary considerably with tectonic setting (VEIZER, 1984; VEIZER and JANSEN, 1979, 1985). For example, the half-life of sediments preserved in platform settings is about 5 times that of sediments deposited in immature erogenic belts and > 10 times that of sediments de- posited in active margin basins and deep sea fans (VEIZER and JANSEN, 1985). The result of these dif- fering rates of recycling is that active margin and im- mature erogenic belt sediments have a greater potential for being lost from the geologic record than are sedi- ments deposited on cratons.

(iii) Erosion and sedimentation producing pelitic Archean sedimentary rocks developed in diverse

rocks, with associated quartzites and carbonates, in- tectonic settings. High-grade supracrustal rocks from dicative of shallow shelf environments. areas such as the Limpopo Province are considered to

(iv) Subsequent burial and granulite-upper am- have been deposited in a stable platform environment

phibolite facies metamorphism to produce metapelites or, possibly, a setting analogous to a modem Passive

and associated metasedimentary rocks. margin (ERIKSSON and KIDD, 1985). The tectonic set-

(v) Uplift to give present exposures. ting of greenstone belt sediments is a matter of con-

Archean REE patterns 2277

siderable debate. but all workers anree that thev were out some of the SSMS analyses, J. M. G. Shelley for the ICP

deposited in tectonically active basins; modern ana- data, Elmer Kiss for the atomic absorption analyses. Bruce

logues are considered to include island arc, back arc Chappell for XRF analyses, Nick Ware for a&stance with

and foredeep basins (DIMROTH et al.. 1982; TARNEY electron probe analyses and Julie Stringer and Mike Shelley forassistance~th the oremration of this oaoer. We are mateftd

et al.. 1976: JACKSON ef al.. 1986). to Peter Gromet, C&in Miller, John P&&al and -Wayne

Regardless of the precise depositional models, it is Sawka for their helpful comments. Fieldwork was supported

clear that greenstone belt environments would be re- by NASA grant NAGW488 to K. A. Eriksson (Limpopo) and

cycled at a rate considerably in excess of cratonic en- the Geological Survey of Western Australia (WGT). We also

vironments preserved in high-grade tetrains such as wish to thank R. Kipling and the Kolokolo bird for the fol- lowing advice:

the Limpopo Province. It follows that these cratonic

the cratonic regions may be over-represented by a factor

environments are preferentially preserved. If relative

of at least 5. This would indicate that such rocks com- ptised no more than about 10% of the exposed Archean

recycling rates from modem environments are adopted,

crust.

“Go to the banks of the

[Kipling, I9021

Editorial handling: Bruce Watson

great grey-green, greasy Limpopo River. . . and find out”

REFERENCES A model for Archean crustal growth

There is substantial evidence that much of the AI&CXE C. J. and BEN D~HMAN D. (I 980) Nd-Sr isotopic

relationship in granitoid rocks and continental crust de- growth and stabilization of the continental crust oc- velopment: a chemical approach to orogenesis. Nature 286,

curred towards the end of the Archean (TAYLOR and 335-342.

MCLENNAN, 1985). However, a major conclusion of BARKER F., ARTH J. G. and HUDSON T. (1981) Tonalites in

the present work is that throughout the Archean, lim- crustal evolution. Phil. Trans. Roy. Sot. London A301,293- uI1 _“<.

ited and isoIatcd areas of crust underwent sub&&al BARTON J. M., RYAN B. and FRIPP R. E. P. (1983) RbSr inttacrustal meltin& formation of upper crustal felsic and U-Th-Pb isotopic studies of the Sand River Gneisscs, rocks depleted in Eu, erosion and deposition of sedi- Central Zone, Limpcpc Mob&e Belt. Ged. SK. &&I Afica

ments depleted in Eu. This process of growth and sta- Spec. Pub!. 8.9-18.

bihzation is directly comparable to the proceszs re- BAMNTON 0. A. and TA~OR S. R. ( 1980) Rare earth element

sponsible for crustaI evolution from the late Archean abundances in Archean metasediments from Kambalda, Western Australia. Geochim. Cosmochim. Acta 44. 639-

onwards. It seems that for most of these areas, subse- 648.

quent burial and metamorphism led to the production BHATLA M. R. (1985) Rare earth element geochemistry of

of high-grade terrains. The reason why these environ- Australian Paleozoic graywackes and mudstones: prove-

ments are almost inv=iabIy P=t=d in Mgh-gradc nana and tectonic setting. Sediment. Geol. 45.97-l 13.

BICKLE M. J. (1978) Heat 10s~ from the cprth: a constraint terrains remains a fundamental question. 0nArchacm~nicsfromthereiationbetweengeotbennal

It is unlikely that these regions comprised more than gtadients and the rate of plate production. Earth Planer.

about 10% of the Archean crust. What were the causes Sci. Left. 40,3OI-315.

of these events? In the Archean, high heat tlow led to Boruc J. L.. DYMEK R. F. and GROMET L. P. (1982) Early

the formation of many small plates (BICUE, 1978). crustal evolution: constraints from variable REE patterns in metasedimentary rocks from the 3800 Ma lsua supra-

Continental areas were also small (TAYLOR and crustal belt, West Gmenland. Lunar Planet. Sci. XIII, 5l- 52. MCLENNAN, 1985). In some cases, tectonic processes,

the nature of which are not well understood, led to thickening (or heating) of the crust with concomitant intracrustal melting. Later, sediments deposited in as- sociated basins were buried to 25-30 km, perhaps by mini-continental collisions with associated overthrust- ing and double thickening of crust. Because only high- grade terrains appear to preserve sediments derived from K-rich granitic sources, the tectonic processes re- sponsible for the development of this type of Archean crust and the later metamorphism may be related. Such environments provide a fundamentally distinct tec- tonic environment from the greenstone belts, which may or may not undergo high-grade metamorphism. Probably crustal growth and stabilization occurred in a sequence of small steplike increments throughout the Archean, with a major pulse of continental growth in the late Archean.

Acknowledgements-We thank Peter Kinny for supplying us with the Mt. Narryer samples, P. Oswald-Scaly for carrying

BRANDL G. (1983) Geology and geochemistry of various su- pmcmsml rocks of the Beit Bridge Complex, east of Measina. Geol. Sec. South Aftica Spec. Publ. 8, 103-l 12.

DANCHIN R. V. ( 1967) Chromium and nickel in the Fig Tree Shale from South Africa. Science 15s. 261-262.

DIMROTH E.. IMREH L., ROCHELMU M. and G~ULFT N. ( 1982) Evolution of the south-central part of the Archean Abitibi Belt, Quebec. Part I: stnttigraphy and pal- graphic model. Can. J. Earzh Sci. 19, 1729-I 758.

DYMEK R. F., WEED R. and GROW L. P. (1983) The Malene metasedimentary rocks on Rypeo. and their relationship to Amitq gneisses. Rapp. GronIands Geol. Unders. 112, 53-69.

ERIKSWN K. A. and KJDD W. S. F. (1985) Sedimentologic and tectonic aspects of the Archean Limpopo Belt (ahstr.). Geol. Sot. Amer. Abstr. Prog. 17,575.

FR~PP R. E. P. (1983) Tbe Precambrian geology of the area around Sand River, near Mesina, Centi Zone, Limpopo Mobile Belt. Geol. Sot. South Afica Spec. Publ. 8.89-102.

FRYER B. J., FYFE W. S. and KERR~CH R. (1979) Archaean volcanogenic oceans. Chem. Geol. 24.25-33.

GEE R. D., BAXTER J. L., WILDE S. A. and WILUAMS I. R. (l98l)Crustal dcntopmmtintheArc&anYilgambIock western Australia. In Archaean Geology(eds. J. E. GLOVER

2278 S. R. Taylor Ed al

and D. I. GROVES), pp. 43-56, Geol. Sot. Aust.. Spec. Publ. 7.

Gt~ns A.K.,Mo~~MERY~.W.,~DAS p.A.andE~s~~v E. A. (1986) The Archean-Proterozoic transition: Evidence horn the geochemistry of metasedimentary rocks of Guyana and Montana. Geochim. Cosmochim. Acta So, 2 125-2 14 I (this issue).

GROMET L. P., DYMEK R. F., HA~KIN L. A. and KOROT!~ R. L. (1984) The “North American shale composite”: Its compilation, major and trace element characteristics. Gee- chim. Cosm~h~m. Acta 48.2469-2482.

HEIER K. S. (1973) Geochemistry of granulite facies rocks and problems of their origin. Phil. Trans. Roy Sot. London A273,429-442.

HORROCK~ P. C. (1983) A corundum and sapphirine para- genesis from the Limpopo Mobile Belt, Southern Africa. J. Metamorph. Geol. 1, 13-23.

JACKSON M. P. A., ERIKSSON K. A. and HARRIS C. W. (1986) Early Archean foredeep sedimentation related to crustal shortening a reinterpretation of the Barberton sequence, southern Africa. Tectonophysics (submitted).

JENNER G. A., FRYER B. J. and MCLENNAN S. M. (198 1) Geochemistry of the Yellowknife Superproup. Geochim. Cosmochim. Acta 45, 111 I- 1129.

KINNY P. D. ( 1986) Zircon ages from the Narryer metamor- phic belt. Eighth Australian Geoi. Conv~t~o~ Abstr. 15, 107.

KIPLING, R. ( 1902) The elephant’s ‘child. && So Stories, Macmillan.

KRONBERG B. I., Ni?swrr H. W. and LAM W. W. (1986) Upper Pleistooene Amazon deep+ea fan muds reflect in- tense chemical weathering of their mountainous source lands. Chem. Geoi. 54,283-294.

MCLENNAN S. M. and TA~OR S. R. (1984) Archaean sedi- mentary rocks and their relatioa to the composition of the

KINNY P. D., WILLLAMS I. S., FRoUDE D. O., IRELAND T. R. and COMPSTON W. (1986) Early Archean zircon ages from orthognciss and anorthosite at Mt. Nartyer, western Australia. Earth Planet. Sci. Lett. (submitt&.

weathering trends of plutonic and volcanic rocks based on thermodynamic and kinetic considerations. Geochrm. c’oc- mochim. Acfn 48, 1.523- 1534.

PERCIVAL J. A. (1983) High-grade metamorphism in the Chapleau-FoJeyet area. Ontario. Amer. Mineral 68,667- 686.

PERCIVAL J. A. and CARD K. D. (1983) Archean crust as revealed in the Kapuskasing uplift. Superior Province, Canada. Geolagv 11,323-326.

PERCIVAL J. A. and KR~GH T. E. ( 1983) U-Pb ztrcon aeo- chrondogy ofthe Kapuskasing str&ur& zone and vi&& in the Chapleau-FoJeyet area. Ontario. Can. J Eurzh Sci. 20, 830-843.

RUDNICK R. L., MCLENNAN S. M. and TAYLOR S. R. (1985) Large ion lithophile elements in rocks from high-pressure granulite facies terrains. Geochim. Cosmochim. Acta 49, 1645-1655.

TARNEY J. and WINDLEY B. F. (1977) Chemistry, thermal gradients and evolution of the lower continental crust. J. Geol. Sue. London 134, 153-172.

TARNEY J.. DALZIEL 1. W. D. and DE WIT, M. J. (1976) Mar- ginal basin ‘Rotas Verdes’ complex from S. Chile: a model for Archaean greenstone belt formation. In The Early His- tory oJthe Earth (ed. B. F. WINDLEY), pp. 13 I - 146. Wiley, London.

TAYLOR S. R. and G~RTON M. P. (1977) Geochemical ap- plication of spark source mass why--fIL Element sensitivity, precision and accuracy. Geochim. Cosmochim. Acta 41, 1375-1380.

312 p. V&tZER J. ( 1984) ReqcJing on the evolving Earth: geocbemical

record in se&men&. Proc. 27th Int. Geol. Congress X1,325- 345.

VELZER J. and JAN~EN S. L. (1979) Basement and sedimentary recycling and continental evolution. J. Geol. 87.341-370.

VELZER J. and JAN~EN S. L. (1985) Basement and sedimentary recycling-2: time dimension to global tectonics. J. Geol.

TAYLOR S. R. and MCLENNAN S. M. (1985) The Continental Crust: Its Composition and Evolution. Blackwell, Oxford,

Arc& amtineatal uust. In Archaean G~h~~t~(~. 93,625-643. A. KRONER, G. N. HANSON and A. M. G00DwtN), pp. WATKEYS M. K. (1983) ~0~~0~ ge&gical map of the 47-72. ~DiiSlner-V&&kZiin. Heidelberg. Limpopo Belt and environs. Geot. Sac. !%uth Ajkica Spec.

McL~ti& S. M., NAN& W. B: and TAYLOR S. R. (1980) Pubi. 8. Rare earth element-thorium correlations in sedimentary WILLIAMS I. R., WALKER I. M., HOCKJNC R. M. and WIL- rocJcs, and the composition of the continental crust. Geo- LV\MS S. J. (1983) Byro, Western Austmha, J:25O,ooO Sheet chim. Cosmochim. Acta 44, 1833-1839. SG50- 10 International Index, Geol. Sur. Western Australia.

MCLENNAN S. M., TAYLOR S. R. and ERIK~~~N K. A. ( 1983) WYLLIE P. J. (1977) Crustal anatexic an experimental review. Geochemistry of Archean shales from the Pilbara Super- Tectonophysics 43.4 t-7 1. IUOUD. Western Australia. Geochim. Cosmochim. Acta 47,

MCLENNAN S. M., TAYLOR S. R. and MCGREGOR V. R. ( 1984) Geochemistry of Arehean metasedimentary rocks from West Greenland. Geochim. Cosmochim. Acta 48, l-

APPENDiX

SAMPLE LOCATIONS AND IMSCRIPTIONS

13. MCLENNAN S. M. MCCULL~CH M. T., TAYLOR S. R. and

MAYNARD 1. B. (1985) Geochemistry of deep sea turbidite sands from difIering tectonic environments (abstr.) EU.S 66, 1136.

MUELLER D. A., WOQDEN J. L. and BOWES D. R. (1982) Precambrian evolution of the Beartooth Mountains, Mon- tana-Wyoming, U.S.A. Rev. Brasil. Geosc. 12.2 15-222.

MYERS J. S. and WILLIAMS I. R. (1985) Earlv Precambrian crustal evolution at Mt. Narryk, we&m k&raJia. Fre- camb. Res. 27, 153-163.

NANCE W. B. and TAYLOR S. R. (1977) Rare earth element patterns and crustal evolution-II. Archean sedimentary rocks from Kalgoorlie, Australia. Geochim. Cosmochim. Acta 41,225-23 1.

NAQVI S. M., CONDIE K. C. and PHILIP A. ( 1983) Geochem- istry of some unusuaJ early Archaean sediments from Dhatwar Craton, India. ~~~r~ Res. 22, 125-147.

NE~B~ H. W. and YOUNO G. M. (J984f Prediction of some

(A) Kapuskasing Structural Zone

KSZ- I pamgneiss. From pragnek unit near kinhoe Lake catachistic zone; within ~o~~xene isopd (PttpCtv~t-, 1983). PJagiocM (An&, garnet and orthopymxenc form porphyroblasts in a matrix of deformed quartz, pkginCJase, aligned biotite, apatite -t zircon. Grthopyroxene is de&tmcd and partially altered to biotite.

Ksz-8 paragneiss. From paragneiss unit near Ivanhoe Lake cataclastic zone; within qx isograd (PERCIVAL, 1983). It COU- s&s of cwwgrained quartz, plagi~lase (A%), blotitc, @r@% graphite, apatite and zircon. Biotite shows preferred orienul- tion; graphite occurs as large plates aligned pataBel to the foliation.

KSZ-I1 paragneiss. From a partiaBy melted pai’rt@ias boulder. on the western end of KS2 within cpx isograd (&a- CIVAL, Ib83). It consists of medium-grained quartz, plaoiodase (An& biotite, game& apatite and zircon; biotite crystals are slightly aligned.

Archean REE patterns 2279

ISZ-I2 and 13 meltedparagneiss. From a single outcrop in the western portion of the KSZ witbin cpx isogmd (PER- CIVAL, 1983). Bothsamplesconsistofquartz,plagioclase(An, in 12 and An= in 13), biotite, apatite and xircon. KSZ-12 is coame&ned with no prefemd orientation of biotite. It con- tains a large fragmented garnet which is Mn-rich (11 wt.%) and shows textural disequilibrium-it is partiahy replaced by biotite and muscovite. KSZ 13 is tine-gained with aligned biotite. Compared with KSZ 12, it has higher proportions of biotite and apatite.

(B) Limpopo Province

LP-4 metapelite. From Constantia farm, 2 km north of Tshipk. Contains porphyroblastic garnet (-5 mm), with aligned inclusions of quartz, occurring in a medium-grained, foliated matrix of quartz, plagioclase, cordierite, biotite, sil- lima&e, opaques and zircon. Retrogressive features include cordierite altered to pink, sericitixed plagioclase and minor chlorite replacing biotite.

LP-I 1 iron-rich metapelite. From Dover farm, southeast of Messina. Medium-gained, polygonal garnet, with rounded quartz inclusions is intergrown with quartz, K-feldspar, cor- dierite and opaques. Minor biotite, zircon and apatite. Feldspar partially sericitixed.

LP-14 f~nous metachert. From Randjesfontein farm, southeast of Messina Layers of magnetite and orthoamphibole are surrounded by quartz, which varies From coarse- to fine- grained. Apatite occurs as an accessory phase. Probably is a chert-rich part of an iron-formation.

U-20 metupefite. From Tweedale farm, north of Alldays. Porphyroblasric plagiocke occurs in foliated matrix of quartz, plagioclase and fibrous biotite. Garnet occurs as minor por- phyroblasts with inclusions of biotite and quartz Zircon and

apatite occur as accessory phases. Phtgioclase is heavily seri- citixed in places, and biotite is partiahy altered to chlorite.

LP-28 metupelite. From Dam Stammetjie se Kop, east of Swartwater, which is west of Alldays. Coame-&ted garnet and quark, with symplectically intergmwn grunerite and pla- @&se. Minor ilmenite and biotite. Plagioclase extensively sericitized.

LP-30 metupehte. From Graat Remet farm, north of Swartwater. Large plagioclase porphyroblasts ate surrounded by foliated matrix of biotite, quartz, cordierite, plegioclase and opaques. Minor, small garnet porphyroblasts occur. Zircon and apatite am accessory phases. Plagioclase heavily sericitixed in places; cot&rite partially altered to pi&e. Minor chlorite alteration developed around biotite.

(C) Western Gneiss Terrain

MN45 metapekte. From pelitic layer within Mt. Narryer sequence, northern Mt. Dugel section. Porphyroblastic garnet and orthopyroxenes are set in a foliated matrix of con&rite, quartz and biotite. Orthopyroxenes have minor alteration rims of biotite and chlorite. Magnetite and zircon are present as accessory phases.

MNK quartz-rich metapelite. From Mt. Narryer sequence. Acicular, aligned sillimanite is surrounded by irregular quartz and cordkite. Accessory phases, identified by electron mi- croprobe, include zircon, rutile, monaxite, xenotime, and a rare Sc silicate, thortveitite (Sc&&). Cordierite variably al- tered to pink.

MNFN metapelite. From Mt. Narryer sequence. Augen of garnet (up to I .5 mm) are surrounded by a foliated matrix of quartz, acicular sillimanite, cordierite, biotite, magnetite and ilmenite. Zircon and monaxite are accessory phases. Cot&rite partially altered in pinite.